[time-nuts] Fury Realhamradio listing

SAIDJACK at aol.com SAIDJACK at aol.com
Mon Aug 20 15:30:12 EDT 2007

In a message dated 8/20/2007 02:26:09 Pacific Daylight Time,  
nebula at telus.net writes:

>A  side effect of the 5 sample average is the opportunity to reduce the  
>effective saw tooth error to 1/5 when the GPS LO is adjusted about 0.4  Hz 
>high. This causes the GPS pulse position to advance about 20ns  every second 
>and repeat every 5 seconds creating a more stable  average.


Hi Peter,
thanks for the info. Not sure you can say 5 sample averages result in 1/5  
the error. I think you have to use 1/sqrt(5) if it's Gaussian noise. Well, a  
hanging bridge is hardly Gaussian, so maybe you are right.
I talked to Paul Allan (Mr. Allan deviation) and Jack Kusters who both  
worked on the Z3816A development, and while they don't remember specifics, they  
did think that the unit has a 1ns TI measurement resolution per second.
It would be interesting to use a highly-stable 1PPS source without sawtooth  
(another Z3816A, or a Rubidium source etc) and replace the GPS receiver 1PPS 
and  see how the algorithms handle this...
The question is: how much short to medium term stability improvement  can be 
achieved with the best available GPS receivers (say an M12M)? Or does it  not 
matter at all as Tom explains, since the loop averages out any STS  errors?
We do see differences in our Fury GPSDO when it is running using an  M12+ 
versus an M12M (all else being the same), the M12M receivers seem to have  
significantly less diurnal pertrubations than the M12+ (well, we are talking  
<+-10ns over a day anyways for the M12+).

************************************** Get a sneak peek of the all-new AOL at 

More information about the time-nuts mailing list